PREPRINT 10-018
CHARACTERIZATION OF DEEP BED FILTER MEDIA FOR
OIL REMOVAL FROM PRODUCED WATER
C. Hank Rawlins, PhD, P.E., eProcess Technologies
Ariane E. Erickson, Montana Tech of the University of Montana
well injection is achieved with deep bed granular media
filters. These filters often use walnut and/or pecan shells,
and are generically called walnut shell filters or nut shell
filters. In addition to produced water treatment, the walnut
shell filter may be used to treat surface waters or brine well
supply waters for solids removal before injection. In all
cases the walnut shell filter serves as the final polishing step
to reduce oil and solids concentration to very low levels and
very fine particle size to protect the disposal or re-injection
well.
The following design boundary conditions are given
for a generic walnut shell filter used in an oilfield
installation. These values can vary widely across various
geographic regions, but serve as a nominal point for process
design.
Water flow rate: 5000-75000 BPD (145-2190 GPM) for a
single vessel
Water temperature: Ambient to 150F
Operating pressure: Atmospheric to 50 psig (3.4 bar)
OIW content: <50 ppm inlet (from deoiler/flotation
outlet), <5 ppm outlet (injection quality)
Solids content: <25 ppm inlet, <5 ppm outlet (injection
quality)
Particle size inlet: <20 micron oil droplets, <10 micron
solids
Particle size outlet: <2 microns oil and solids
Media: Walnut shell (primary), pecan shell (secondary)
A flow schematic of a vertical walnut shell filter is
shown in Figure 1. While there are several patented or
proprietary designs offered by walnut shell manufacturers,
the flow schematic shown is for illustration purposes only.
It is loosely based on US patent 4,826,609. Water with oil
and solids enters near the top of the vessel through a
distribution nozzle. The bulk flow of water is downward,
where ideally a plug flow regime of the water flows through
a bed of granular filter media. This media may be 36-72
in depth and span the full diameter of the vessel. The
nominal flux rate is 10-16 gpm/ft. The media is retained by
a screen, which allows the cleaned water to flow through
and report out the bottom of the vessel. Initial pressure drop
across the bed is approximately 1 psi, and as the differential
climbs to 15-20 psi, a backwash cycle is initiated. During
Abstract
Deep bed media filtration is used as a tertiary treatment
step (after hydrocyclones and flotation cells) to remove fine
oil droplets and solids from oilfield water before reinjection
disposal or to remove particulate material from surface
waters used for waterflood injection. Oil and solids
injection requirements are typically <5 ppm concentration
at <2 micron diameter. The preferred media for produced
water filtration is walnut or pecan shell, however the use of
this media has evolved by trial & error and the fundamental
mechanisms that occur during separation and media
cleaning have not been investigated. A project was initiated
to fill in this knowledge gap by analyzing the physical,
mechanical, chemical, and hydrodynamic properties of six
types of granular filtration media (12/20 mesh size): garnet,
silica, anthracite, pecan shell, English walnut shell, and
black walnut shell. Comparison of bed packing, oil wetting,
attrition resistance, particle strength, and fluidization has
led to an understanding of the preferred filtration media for
separation of fine oil droplets and solids.
Water
Inlet
Backwash
Pump &
Screen
Bulk
Water
Flow
Backwash
Outlet
Granular
Filtration
Media
Retention
Screen
Filtrate
100.0
90.0
80.0
70.0
60.0
50.0
English
40.0
Pecan
30.0
Black
Silica
20.0
Anthracite
10.0
Garnet
0.0
600
800
1000
1200
1400
1600
1800
2000
2200
2400
Bulk Density
A glass 1000 ml graduated cylinder was selected and
the weight recorded. Approximately 100 ml of media
material (grab sample) was added and the cylinder tapped
10 times on the flat portion of the cylinder. The process was
repeated with 100 ml of material and 10 taps until 900 ml
was reached. The weight of the cylinder with dry material
was recorded to determine dry bulk density and open
volume % dry (using apparent density). Using a large
burette, tap water was added until reaching the level of the
media, and the amount of water (ml) was recorded along
with the weight of the cylinder (media plus water) and
resulting level of water/media mixture (height may change
due to absorption swelling). The cylinder was then inverted
onto a 600 micron (U.S. 30 mesh) sieve, and the media and
water removed. The free water was allowed to drain
through the sieve. The net wet weight of the media was
recorded to determine the water retention quotient (i.e.
adsorbed + absorbed water in g water/g media). All weights
were recorded to 0.00 grams accuracy. The results are
shown in Table 2.
The dry bulk density followed the same trend as
apparent density, with the biological materials exhibiting
the lowest values, and garnet the highest values. The dry
open volume % though was nearly the same, regardless of
material type. This is because of the narrow size
distribution leading to similar packing efficiency. The water
retention quotient is a combination of absorbed water into
the pores of the granular media plus adsorbed water on the
surface. Garnet, followed by silica, had the lowest water
retention due to low surface absorption. Of the biological
materials the water retention quotient approximately
followed the % void space (Table 1) with black walnut
having the lowest amount of water absorbed.
15.97
12.61
11.61
10.20
6.08
3.05
Garnet
Silica
Black
Anthracite
English
Pecan
0.70
Dodecane
0.50
Mobil
Tectyl
0.40
0.30
0.20
0.10
Density
@ 22C
(g/cm)
0.747
0.848
0.916
API
Gravity
57.9
35.4
23.0
Visc.
@ 22C
(cP)
1.3
15.9
292.4
0.00
Surf.
Tens.
(dyne/cm)
38.2
29.7
32.5
Black
English
Pecan
Anthracite
Garnet
Silica
For dry media testing, approximately 25 grams of 1020 mesh sized material was weighed out and placed into a
clean pan. The media was covered completely with the test
oil for 20 minutes. The mixture was then poured on a 25
mesh screen, spread into a monolayer, and allowed to stand
for 5 minutes while oil drains by gravity. The oily media
was removed from the screen and weighed on filter paper.
The procedure was repeated for the three test oils. For wet
media testing, approximately 25 grams of the 10-20 mesh
material was weighed out, placed in a beaker, and
completely submerged with tap water for one hour. The
excess water was removed and the media filtered through a
vacuum filter for five minutes to remove any free water.
The wet media was transferred onto a dry piece of filter
paper to record the saturated media weight. The wet media
was then subjected to the same oil immersion test as the dry
media. All weights were recorded to 0.00 grams accuracy.
The results are shown in Figures 4 and 5.
The three test oils were selected to provide a range of
API densities. Dodecane is a light condensate with 58 API
gravity, Mobil Velocite #6 is a medium weight oil with 35
API gravity, while Tectyl is a heavy oil with 23 API
gravity. The dry silica and garnet media showed
0.60
Dodecane
0.50
Mobil
Tectyl
0.40
0.30
0.20
0.10
0.00
Black
English
Pecan
Anthracite
Garnet
Silica
Pecan
Silica
1.5
7.0
English
Garnet
1.0
0.5
Black
6.0
English
Black
2.0
Pecan
Silica
5.0
Garnet
0.0
Anthracite
4.0
10
20
30
40
50
60
3.0
2.0
1.0
0.0
0
10
20
30
40
50
60
Fluidization
For the silica and garnet tests, approximately 48 inches
of dry media was measured out. For the anthracite, black
walnut, English walnut, and pecan, approximately 41 inches
of dry media was measured out and soaked in tap water for
one hour. All media was then added to the column,
fluidized with tap water, and allowed to settle to determine
Self-attrition index is a representation of the particleparticle interaction that occurs during fluidization and
backwash. Ignoring interaction with the vessel walls, most
of the particle wear arises from particle-to-particle impact
during these events. This test was to provide an index of the
Silica
Garnet
60.0
Anthracite
Pecan
English
50.0
70.0
Drain
4 X 1
8 10
106 H x 4 D
Column
Black
40.0
30.0
20.0
10.0
4 10
10
0.0
0.0
Media Bed
PI
10.0
15.0
20.0
25.0
30.0
35.0
40.0
24 Mesh
Screen
8.0
4 Mesh
Screen
FI
5.0
7.0
4 X 1
6.0
Water
Supply
5.0
Silica
Garnet
4.0
English
Black
3.0
Pecan
2.0
1.0
0.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
Water
Supply
FI
4 X 1
8 10
106 H x 4 D
Column
4 10
Pressure/Flow Relationship
The pressure required to achieve a specific flux rate is
a key sizing parameter of a deep bed media filter. The flux
rate versus pressure drop for the six granular media was
measured using a 4 inch column. Each media was presoaked in water for at least one hour then a specific height
measured into the 4 inch diameter column. Each media was
tested at 24 inch, 48 inch, and 72 inch bed height. Water
was introduced into the top of the column at a measured
flow rate, and the resulting pressure drop determined. For
each test the pressure drop was measured as the head of
water above the media bed. As the media was suspended on
a screen to allow free space below the bed (e.g. no back
pressure from piping), the height of water measured above
the bed will give the pressure drop across the specific bed
depth.
Figure 11 shows a schematic of the test apparatus built
and used for measuring the pressure/flow relationship. The
main column was built of 4 inch schedule 40 clear PVC and
uses 24 mesh and 4 mesh (304 SS) screens to hold the
media. Water supply and drains from the column were
through inch garden hose, while all 1inch lines were gray
PVC. The flow control valve was a manual globe valve
(class 125, 1 inch) and the flow indicator was a digital
flowmeter/totalizer with 1 inch nylon body and 0.3-3 GPM
capacity.
Figure 12 shows the water flux (gpm/ft) versus
pressure drop (psi) for six media with a nominal 72 inch
bed height. The pressure drop for all media and all flow
rates tested was very low. Within the normal flux range
experienced (12-16 gpm/ft) the pressure drop for all media
was <1 psi. Silica and anthracite exhibited a slightly higher
pressure drop because of particle shape and size. These two
media types were flat and angular and had the smallest
particle size, and when packed together in a bed increased
the resistance to flow. The three biological media (English
walnut, black walnut, and pecan) all exhibited very similar
pressure drop characteristics. The pressure drop for all
media showed a linear relationship with flux.
Media Bed
24 Mesh
Screen
10
4 Mesh
Screen
4 X 1
Drain
0.70
English Walnut
Black
0.60
Pecan
Anthracite
0.50
0.40
0.30
0.20
0.10
10.0
15.0
20.0
25.0
Figure 12. Pressure drop versus filtration flux for all six
media types.
4.
5.
6.
7.
References
1.
2.
3.
10
Evans, P., K. Robinson. Produced water managementreservoir and facilities engineering aspects. SPE
Paper 53245 Presented at the 1999 SPE Middle East
Oil Show, Bahrain, 20-23 February 1999.
Mantilla, I., P. Quintero. Comprehensive system for
treatment and injection of produced water: field cases.
SPE Paper 63169 Presented at the 2000 SPE Annual
Technical Conference and Exhibition, Dallas, TX, 1-4
October 2000.
Blumenschein, C.D. Walnut shell filtration for oil and
solids removal from steel mill recycle systems. AISE
Steel Technology 78.4 (2001): 33-37.
Srinivasan, A., T. Viraraghavan. Removal of oil by
walnut shell media. Bioresource Technology 99
(2008): 8217-8220.